Layout Analysis Method With Antecedent Non_text Regions

Layout analysis is an important part in document layout analysis and understanding. It is used to transfer content in paper document to electronic digital information for further digitalization of total layout. The accuracy of the layout analysis has a direct impact on the results of understanding and determines whether the semantic relations and logical relationship is correct in the output of the layout of information processing system.Thus,the study of layout analysis is of important theoretical significance and application value.As the non-text regions will disturb the pick-up of the text regions,we provide a layout analysis method with antecedent non-text regions for this characteristic. First we detect layout skew based on window transform,then extract the non-text regions and remove them to avoid the disturbance to the pick-up of the text regions. This paper applies the methods based on projection,run-length smoothing and minimal spanning tree lustering to process the non-embedded rectangular layout and the embedded one. At last it determines the logical order in layout based on directed graph algorithm.It can infer from experiments that the mothod is better to segment the rectangular layout.